Channel reflection: Knowledge-driven data augmentation for EEG-based brain-computer interfaces.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Brain-computer interfaces (BCIs) facilitate direct brain-device communication.
• Electroencephalography (EEG)-based BCIs are popular but face calibration data limitations.
• Developing robust decoding models is challenging with scarce user-specific EEG data.

Purpose of the Study:

• To address the challenge of limited calibration data in EEG-based BCIs.
• To propose a novel parameter-free data augmentation technique called channel reflection (CR).
• To evaluate the effectiveness and robustness of CR across diverse BCI paradigms.

Main Methods:

• Developed a parameter-free channel reflection (CR) data augmentation approach.
• Incorporated prior knowledge of channel distributions specific to BCI paradigms.
• Tested CR on eight public EEG datasets spanning four BCI paradigms (motor imagery, SSVEP, P300, seizure classification).
• Evaluated CR performance with various decoding algorithms.

Main Results:

• Channel reflection (CR) significantly improves classification accuracy in EEG-BCIs.
• CR demonstrates superior performance compared to existing data augmentation techniques.
• CR is flexible and can be integrated with other augmentation methods for enhanced results.

Conclusions:

• Data augmentation, particularly CR, is crucial for advancing EEG-based BCIs.
• CR offers an effective, robust, and flexible solution to data scarcity in BCI calibration.
• The proposed CR method should be considered a standard component in EEG-BCI development.